path: root/Master/tlpkg/bin/tlpkg-ctan-check

#!/usr/bin/env perl
# $Id$
# Public domain.  Originally written 2005, Karl Berry.
# Rewritten for new infrastructure 2007, Norbert Preining.
# Check if a package in TL has any changes on CTAN.  Just the beginnings.

BEGIN {
  chomp ($mydir = `dirname $0`);  # we are in Master/tlpkg/bin
  unshift (@INC, "$mydir/..");
}

use TeXLive::TLConfig;
use TeXLive::TLPOBJ;
use TeXLive::TLPDB;

use File::Basename;

my $tlpdb;
my $Master;
our %OPT;

my @WorkingTLP = qw(
        12many
        a0poster a2ping aastex abbr abc abstract abstyles accfonts achemso
          acmconf acmtrans acronym 
          active-conf addlines adrconv ae afthesis
          aguplus aichej AkkTeX akletter
          algorithm2e algorithmicx algorithms alnumsec altfont ametsoc ams
          amsaddr amslatex-primer amsldoc-it amsmath amsrefs
          amsthdoc-it animate ANUfinalexam anyfontsize anysize
          apa apacite appendix apl
          arabi arabtex arabxetex archaic arcs arev
          armenian AroBend arphic arrayjob arydshln Asana-Math ascii
          assignment astro
          attachfile auncial-new aurical authoraftertitle authorindex
          auto-pst-pdf autoarea automata
        babel babelbib bangtex barcodes bbcard bbold
          beamer beamer-contrib beamer-tut-pt begriff bera betababel beton
          bez123 bezos bibhtml bibleref bibtopic bibtopicprefix
          bibunits bidi bigfoot binomexp blacklettert1 blindtext
          blockdraw_mp blowup boisik
          boites boldtensors bookest booklet booktabs bophook
          boxhandler bpchem breakurl bullcntr bussproofs bytefield 
        captcont caption carlisle catechis carolmin-ps
          cbcoptic cbfonts ccaption cclicenses cd cdpbundl cellspace
          changebar changepage changes chappg
          chem-journal chemcompounds
          chemstyle cherokee chessboard chessfss chletter circ
          citeref cjhebrew cjk classicthesis
          clefval cleveref
          clock clrscode cm-super
          cmap cmastro cmcyr cmdstring cmdtrack cmll colorinfo
          colortbl colorwav commath compact compactbib complexity comprehensive
          computational-complexity confproc constants
          context-account context-bnf context-chromato 
          context-construction-plan context-degrade context-french
          context-letter context-lettrine context-lilypond context-mathsets
          context-taspresent
          context-typearea context-vim
          cooking cool coollist coolstr cooltooltips
          coordsys courier-scaled courseoutline coursepaper coverpage
          crop crossreference cryst csbulletin csquotes csvtools ctable
          cursolatex cursor cuisine
          currvita curve curve2e curves
          custom-bib cv cweb-latex cyklop cyrillic
        dancers dashbox dashrule datatool
          dateiliste datetime dcpic decimal dehyph-exptl
          delimtxt digiconfigs din1505
          dinat dingbat directory disser dk-bib dlfltxb docmfp doi doipubmed
          dotarrow dottex
          doublestroke dpfloat dprogress drac draftcopy
          draftwatermark dramatist dtxgallery
          dtxtut duerer-latex dvdcoll dviasm dviincl dvipdfmx-def
          dvipsconfig dyntree
        ean13isbn easy ebong ebsthesis eCards ecc ecclesiastic economic
          ecv ed edmac edmargin ednotes eemeir egameps
          egplot eijkhout
          ellipsis elpres elsarticle elsevier elsevier-bib
          emulateapj emp encxvlna endfloat endheads
          engpron engrec enumitem envbig environ envlab
          epigrafica epigraph epiolmec eplain
          epsdice epsf epsincl epslatex-fr epspdfconversion
          eqlist eqnarray eqparbox errata es-tex-faq
          ESIEEcv esdiff esint esint-type1 eskd eskdx eso-pic esvect etaremune
          etex-pkg ethiop ethiop-t1 etoolbox euenc eukdate
          eulervm euproposal euro europecv eurosans everypage exam
          examdesign examplep exercise expdlist expl3 expressg extarrows exteps
          extpfeil extract extsizes
        faktor fancybox fancyhdr fancynum fancyref fancytooltips fancyvrb
          FAQ-en fax
          fbs featpost
          feyn feynmf fge filecontents fink fixme
          flagderiv flippdf float floatrow flowfram fltpoint
          fmtcount fnbreak
          fncychap foekfont fontch fontspec fonttable fontwrap
          footbib footmisc forarray forloop formular
          fouridx fourier fouriernc
          fp fpl frankenstein free-math-font-survey
          frenchle fribrief frletter frontespizio fullpict functan
        gaceta galois gastex gatech-thesis gb4e gcard gcite
          genmisc genmpage gentle germbib
          gfsartemisia gfsbodoni gfscomplutum gfsdidot gfsneohellenic
          gfsporson gfssolomos
          ginpenc gloss glossaries gmdoc gmeometric
          gmiflink gmutils gmverb gnuplottex gost graphics
          graphicx-psmin greek-inputenc greekdates greenpoint
          grnumalt gu GuIT guitar guitbeamer
        HA-prosper hands hanging harmony hep hepnames
          hepparticles hepthesis hepunits hypdvips
        hhtensor histogr hitec hrlatex hvfloat 
          hyperref hyperxmp hyph-utf8 hyphen-base 
          hyphenex hyplain
        ibygrk icsv IEEEconf IEEEtran ifplatform ifxetex
          ijmart ijqc image-gallery impatient impatient-fr imtekda
          inlinebib inlinedef
          interactiveworkbook inversepath invoice iopart-num iso
          iso10303 isodate isodoc isonums isorot isotope itnumpar iwona
        jeopardy jknapltx jneurosci jpsj juraabbrev jurabib juramisc jurarsp
        kalender kastrup kerkis kerntest keystroke knuth koma-script kpfonts
        l2picfaq l2tabu-english l2tabu-spanish
          labbook labelcas labels lastpage
          latex latex2e-help-texinfo latex-course latex-tds latexmp
          layouts lcd lcg lcyw leading leaflet ledmac leftidx lettrine
          lewis lfb lgreek lh
          libertine limap linearA lineno linguex
          lipsum listings listliketab lkproof lm lnotes
          logpap lps lsc lshort-bulgarian lshort-chinese
          lshort-dutch lshort-english
          lshort-finnish lshort-french lshort-german lshort-italian
          lshort-japanese lshort-korean lshort-mongolian
          lshort-polish lshort-portuguese lshort-slovak lshort-slovenian
          lshort-spanish lshort-thai lshort-turkish lshort-ukr
          lshort-vietnamese ltablex ltabptch ltxindex ltxmisc lxfonts
        mafr magyar mailing makebox makecell makecirc makeglos
          makebarcode makeplot maple marginnote marvosym mathdots mathexam
          mathmode maybemath
          mcaption mceinleger mciteplus mdwtools memexsupp
          memoir MemoirChapStyles mentis
          menu metaobj metaplot metatex metauml mff mfpic mftinc mh mhchem
          mhequ mhs microtype midnight minipage-marginpar minitoc minutes
          mkjobtexmf mkpattern
          mla-paper mlist mmap mnsymbol
          moderncv modroman mongolian-babel moreverb movie15 mp3d mparhack
          mpman-ru ms msc msg multicap
          multido munich muthesis mwcls
        nag namespc natbib nath nature ncclatex ncctools newlfm nih noitcrul
          nomencl nomentbl nostarch notes2bib
          ntgclass ntheorem-vn numline numname numprint
        oberdiek ocr-latex octavo ofs ogham oldstyle onlyamsmath opcit
          ordinalpt
        pagenote paper papercdcase papertex parallel paresse parrun passivetex 
          patch pauldoc pbox pdf-forms-tutorial-de pdf-forms-tutorial-en
          pdf-trans pdfcprot pdfcrop
          pdfpages pdfscreen pdfsync pdftex-def pdftricks perception perltex
          permute petiteannonce
          pgf pgf-soroban pgfopts pgfplots phaistos philex philosophersimprint
          pict2e pictex pictex2 pinlabel pittetd placeins plari plates play
          plnfss poemscol polski polyglossia polynom polynomial polytable
          powerdot ppr-prv pracjourn preprint
          prettyref preview proba probsoln procIAGssymp program protex protocol
          pseudocode psfrag psfrag-italian psnfss pspicture
          pst-2dplot pst-3d pst-3dplot pst-asr pst-bar
          pst-barcode pst-blur pst-circ pst-cox pst-dbicons pst-diffraction
          pst-eps pst-eucl pst-fill
          pst-fr3d pst-fractal pst-fun pst-func
          pst-geo pst-grad pst-infixplot pst-jtree pst-labo
          pst-lens pst-light3d pst-math pst-ob3d pst-optexp pst-optic
          pst-osci pst-pad pst-pdf pst-pdgr pst-qtree
          pst-solides3d pst-soroban pst-spectra
          pst-slpe pst-stru pst-uml pst-vue3d
          pst2pdf pstricks pstricks-add
          ptptex pxfonts
        qcm qstest quotmark qtree
        r_und_s randbild rccol rcs rcsinfo recipecard
          refstyle regcount register resume robustcommand
          robustindex rotating rsc rtkinenc rtklage
        sagetex sauerj sauterfonts savefnmark savesym savetrees scale
          scalebar schedule scientificpaper sciposter screenplay script
          sdrt sectionbox sectsty semantic semaphor seminar semioneside
          seqsplit setspace sf298 sffms sfg sfmath sgame shadbox shapepar
          shipunov shorttoc show2e showdim showexpl showlabels
          SIstyle SIunits
          sidecap sides siggraph simplecv simplewick siunitx
          skak skaknew slideshow smalltableof smartref
          snapshot songbook sort-by-letters soul spanish-mx sparklines spie
          splines splitbib splitindex spotcolor sprite srcltx sseq
          ssqquote stack stage statex2 staves stdclsdv stdpage stellenbosch
          stex stringstrings struktex
          sttools stubs subdepth subeqn subeqnarray
          subfig subfigure subfloat substr sudoku sudokubundle sugconf 
	  supertabular susy svgcolor svn svn-multi svninfo
          swebib syntax synproof syntrace synttree
        t-angles Tabbing tableaux tablists tablor tabulary tabvar talk 
	  tamethebeast tdsfrmath
          ted templates-fenn templates-sommer tengwarscript
          tensor teubner tex-gyre texilikecover texmate
          texpower texshade
          textcase textfit textopo textpath textpos
          theoremref thesis-titlepage-fhac
          thinsp thmtools thumb thumbpdf thuthesis ticket tikz-inet
          titlefoot titlesec
          titling tocbibind
          tocloft todo tokenizer toolbox tools toptesi totpages tracking
          trajan translator tree-dvips trfsigns trivfloat trsym
          tufte-latex tugboat tugboat-plain turnstile twoup txfonts Type1fonts
          typedref typogrid
        uebungsblatt ucthesis uhrzeit uiucthesis ulsy umich-thesis uml umlaute
          umthesis umtypewriter underlin undertilde units unitsdef unroman
          upmethodology upquote urlbst ushort uwthesis
        vancouver variations varindex varsfromjobname
          vector velthuis venn verse versions vertex
          vhistory visualfaq vmargin vntex volumes vpe vwcol vxu
        wadalab wallpaper warning warpcol williams wordlike wrapfig
        xargs xcolor xdoc xecyr xepersian
          xetex-def xetex-pstricks xetexfontinfo
          xetexref xfor xgreek xifthen
          xkeyval xltxtra xmpincl xnewcommand xoptarg xpackages
          xq xskak xstring xtab xtcapts xunicode xyling xypic-tut-pt xytree
        york-thesis
        ziffer zwgetfdate
    ); 


exit (&main ());

sub main
{
  # get the Master dir and texlive.tlpdb
  chomp ($Master = `cd $mydir/../.. && pwd`);
  $tlpdb = TeXLive::TLPDB->new ("root" => $Master);
  die "Cannot find tlpdb in $Master!" unless defined $tlpdb;

  $OPT{"verbose"} = 0;
  if ($ARGV[0] eq "--verbose") {
    $OPT{"verbose"} = 1;
    shift @ARGV;
  }
  
  if ($ARGV[0] eq "--list-not-treated") {
    print "TeX Live packages not in $0:\n";
    for my $b (&normal_tlps ()) {
      print "$b\n" if ! grep (/^$b$/, @WorkingTLP);
    }
    @ARGV = (); # no normal checks

  } elsif ($ARGV[0] eq "--check") {
    # check all/only those packages we have actually run through this mill.
    @ARGV = @WorkingTLP;
  } elsif ($ARGV[0] eq "--check-all") {
    @ARGV = &normal_tlps ();
  }

  my $errcount = 0;
  for my $tlp (@ARGV) {
    print "checking $tlp..." if $OPT{"verbose"};
    $errcount += &do_tlp ($tlp);
  }
  return $errcount;
}

# gives a list with only the 'normal' packages,
# that is, excluding meta-packages and binary packages
# (and hyphen for the moment)
# 
sub normal_tlps
{
  my @normal;
  my $non_normal = `ls $Master/bin`;
  $non_normal =~ s/\n/\$|/g;
  $non_normal .= '^0+texlive|^bin-|^collection-|^scheme-|^texlive-|^hyphen-';
  foreach ($tlpdb->list_packages) {
    push (@normal, $_) unless (/$non_normal/);
  }
  return @normal;
}


# Return 1 if package needs updating, 0 if ok.
# 
sub do_tlp
{
  my ($tlpn) = @_;
  my $needed = 0;
  
  my $tlp = $tlpdb->get_package($tlpn);
  if (not(defined($tlp))) {
    warn "$0: no package $tlpn\n";
    return 1;
  }

  chomp (my $ctan_dir = `$mydir/tlpkginfo --prepare '$tlpn'`);
  if (! $ctan_dir) {
    warn "$0: oops, no CTAN directory for $tlpn, fix fix\n";
    return ();
  }
  my @tpm_files = ();
  push @tpm_files, $tlp->runfiles;
  push @tpm_files, $tlp->docfiles;
  push @tpm_files, $tlp->srcfiles;
  # we don't push bin files, should we?
  
  my @compared = ();
  for my $file (@tpm_files) {
#warn "checking file $file\n";
    my $tl_file = "$Master/$file";
    if (! -e $tl_file) {
      warn "$tl_file: TL file missing\n";
      next;
    }
    
    (my $basefile = $file) =~ s,^.*/,,;
    chomp (my @ctan_files = `find $ctan_dir/ -name $basefile`);
#warn "ctan find $basefile: @ctan_files\n";
    # the trailing / is so if we happen to hit a symlink on CTAN, it'll
    # go through.
    next if @ctan_files > 1; # if more than one file by same name, skip

    my $ctan_file = $ctan_files[0];
#warn "ctan file is $ctan_file\n";
    if (! -e $ctan_file) {
      # maybe it'll be there with a case change in the name
      chomp (@ctan_files = `find $ctan_dir/ -iname $basefile`);
      next if @ctan_files > 1; # if more than one file by same name, skip
      $ctan_file = $ctan_files[0];
      
      if (! -e $ctan_file) {
        # we generate lots of files, eg perlmacros.sty, so might skip.
        warn "$ctan_file: CTAN file missing\n"
        if $ctan_file && $ctan_file !~ /(cfg|dvi|sty|tex)$/;
        next;
      }
    }

    push (@compared, $basefile);
    if (&files_differ ($tl_file, $ctan_file)) {
      # we sometimes regenerate pdf's for TL, so we don't complain if
      # they are different.  On the other hand, we don't always
      # regenerate them, so might as well check first.
      next if $tl_file =~ /\.pdf$/;

      print "# $tlpn\ndiff $ctan_file $tl_file\n";
      $needed = 1;
      last;
    }
  }
  
  # xx check ctan dir for new files
  
  if (@compared == 0) {
    warn "\n$tlpn: no files to compare in $ctan_dir, fixme!\n";
    warn "(tpm_files = @tpm_files)\n";
  } elsif ($needed == 0) {
    print "ok, " . (@compared + 0) . " compared (@compared)\n"
      if $OPT{"verbose"};
  }

  # clean up the tmpdir possibly created when invoking 'tlpkginfo --prepare'
  chomp (my $ctan_root = `tlpkginfo --ctan-root`);
  system ("rm -rf $ctan_dir") unless ($ctan_dir =~ m#^$ctan_root#);

  return $needed;
}



# 0 if files are the same, 1 if they are different.
# 
sub files_differ
{
  my ($tl_file,$ctan_file) = @_;
#warn "comparing $ctan_file $tl_file\n";
  return system ("cmp-textfiles $ctan_file $tl_file");
}

# vim: set ts=8 sw=2 expandtab: